Accurate and complete human pathway information is crucial for the successful exploitation of the human genome in a variety of experimental contexts. The Reactome project has enjoyed a successful pilot phase to develop appropriate data models and tools for storing pathway information. In the pilot we have used these tools to curate to the highest level of accuracy pathways involving 1,200 non redundant proteins, which compares favourably to other pathway databases. In this proposal we aim to populate Reactome with pathways which comprise 5,000 non redundant human proteins and extend Reactome to incorporate an additional 10,000 proteins. Consistent with the pilot, all of Reactome's data and software will be made freely available.